QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Vol. 19, Issue 2, 523-535, February 2008

This Article Abstract Full Text (PDF) Supplemental Materials All Versions of this Article: E07-04-0394v1 19/2/523    most recent Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Manolea, F. Articles by Melançon, P. PubMed PubMed Citation Articles by Manolea, F. Articles by Melançon, P.

Distinct Functions for Arf Guanine Nucleotide Exchange Factors at the Golgi Complex: GBF1 and BIGs Are Required for Assembly and Maintenance of the Golgi Stack and trans-Golgi Network, Respectively

Florin Manolea^*, Alejandro Claude, Justin Chun^*, Javier Rosas, and Paul Melançon^*

*Department of Cell Biology, University of Alberta, Edmonton, AB, T6G 2H7 Canada; and Instituto de Bioquímica, Universidad Austral de Chile, Campus Isla Teja, Casilla 567, Valdivia, Chile

Submitted May 1, 2007; Revised October 31, 2007; Accepted November 6, 2007
Monitoring Editor: Benjamin Glick

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
We examined the relative function of the two classes of guanine nucleotide exchange factors (GEFs) for ADP-ribosylation factors that regulate recruitment of coat proteins on the Golgi complex. Complementary overexpression and RNA-based knockdown approaches established that GBF1 regulates COPI recruitment on cis-Golgi compartments, whereas BIGs appear specialized for adaptor proteins on the trans-Golgi. Knockdown of GBF1 and/or COPI did not prevent export of VSVGtsO45 from the endoplasmic reticulum (ER), but caused its accumulation into peripheral vesiculotubular clusters. In contrast, knockdown of BIG1 and BIG2 caused loss of clathrin adaptor proteins and redistribution of several TGN markers, but had no impact on COPI and several Golgi markers. Surprisingly, brefeldin A–inhibited guanine nucleotide exchange factors (BIGs) knockdown prevented neither traffic of VSVGtsO45 to the plasma membrane nor assembly of a polarized Golgi stack. Our observations indicate that COPII is the only coat required for sorting and export from the ER exit sites, whereas GBF1 but not BIGs, is required for COPI recruitment, Golgi subcompartmentalization, and cargo progression to the cell surface.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The Golgi complex, the central sorting station of the secretory pathway, exhibits a very intricate organization that likely reflects the complexity of trafficking and sorting events that take place within this organelle (Rambourg and Clermont, 1990; Mogelsvang et al., 2004). It comprises three main structural elements: two extensive tubular-reticular networks flanking the cis- and trans-sides of a central stack of flat disk-shaped cisternae. The flat cisternae that form the core of the Golgi complex (compact zones) appear interconnected by lateral tubular networks (noncompact zones) to form a continuous membrane ribbon (Rambourg and Clermont, 1990; Mogelsvang et al., 2004). Subcellular fractionation and immunocytochemical analysis further established that cis-, medial, and trans-Golgi elements contain different sets of resident enzymes and that the Golgi complex is therefore functionally compartmentalized (Farquhar and Palade, 1998; Polishchuk and Mironov, 2004). For example, the early acting enzyme mannosidase I localizes to cis-Golgi membranes, whereas later acting ones such mannosidase II (ManII) and sialyltransferase concentrate in medial and trans compartments, respectively.

Imaging of live cells revealed that the Golgi complex is not static as initially assumed from its intricate structure, but rather surprisingly dynamic and linked to several other organelles by active bidirectional transport routes (Bonifacino and Glick, 2004). In animal cells, cargo initially translocated into the endoplasmic reticulum (ER) is selected for transport from specialized ER exit sites (ERES) to vesiculotubular clusters (VTCs; Fromme and Schekman, 2005; Tang et al., 2005). Shortly after their formation, these pleiomorphic carriers are transported on microtubules toward the Golgi complex (Presley et al., 1997; Scales et al., 1997) where they collect, fuse into a network, and subsequently become a flattened cis-Golgi cisterna (Bonifacino and Glick, 2004); peripheral VTCs and those accumulated near the Golgi are collectively called ER-Golgi intermediate compartment or (ERGIC; Appenzeller-Herzog and Hauri, 2006). Current evidence suggests that cargo molecules then progress through the Golgi stack by a process termed cisternal maturation whereby newly formed cis-cisternae containing cargo progressively move toward the trans side as they lose early acting Golgi enzymes and acquire late acting ones (Puthenveedu and Linstedt, 2005; Losev et al., 2006; Matsuura-Tokita et al., 2006). At the trans-Golgi network (TGN), cargo is then sorted to destinations that include the endosome, plasma membrane (PM), lysosomes, or secretory granules (Rodriguez-Boulan and Musch, 2005).

Formation of cargo carriers depends on the spatially and temporally regulated membrane recruitment of specific coat proteins (COPs) from the cytoplasm. At the membrane, COPs select cargo and serve as a scaffold for membrane deformation and vesicle budding (Bonifacino and Glick, 2004; Rabouille and Klumperman, 2005). Transport between the ER and the Golgi complex involves two types of COPs. The COPI coat, first identified in situ at the periphery of the Golgi (Orci et al., 1986), has been implicated in both anterograde and retrograde traffic between the Golgi and VTCs (Duden, 2003). COPII-coated structures on the other hand, form at ERES and mediate export of cargo from the ER (Barlowe, 2003; Tang et al., 2005). Packaging of endosome-targeted cargo at the TGN involves clathrin and several adaptor proteins (AP), including the multimeric AP-1, AP-3, and AP-4 and the monomeric gamma ear Golgi-localized Arf-binding protein (GGAs; Bonifacino, 2004; Robinson, 2004). Current evidence suggests a multistep process in which all three GGAs act in concert at the TGN to concentrate their ligands in coated regions for eventual transfer to AP-1 (Ghosh and Kornfeld, 2004). The recently identified exomer may play similar function for specialized endosomal cargo (Wang et al., 2006).

The recruitment of COPs and their adaptors is controlled by small GTPases, which in turn are regulated by guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (D'Souza-Schorey and Chavrier, 2006; Nie and Randazzo, 2006). Sar1 initiates recruitment of COPII (Sato and Nakano, 2007) whereas ADP-ribosylation factors (Arfs) regulate assembly of not only COPI, but also AP-1, -3, and -4 and the GGAs (D'Souza-Schorey and Chavrier, 2006). Arfs have been divided into three classes based on sequence similarity. Arf1 and Arf3, members of the class I (Arf1-3), as well as the class II Arf5, have been localized to the Golgi complex; in contrast, Arf6, the only class III Arf, associates primarily with the plasma membrane and endosomes (Donaldson and Honda, 2005). Several Arf-GEFs have been characterized, all of which contain a conserved Sec7 domain responsible for Arf activation (Cox et al., 2004; Mouratou et al., 2005). Two Arf-GEF subfamilies called GBF1 and BIGs appear to regulate Arf activation for coat recruitment on the Golgi complex. Both of these GEFs are inhibited by the fungal metabolite brefeldin A (BFA; Mansour et al., 1999; Togawa et al., 1999; Niu et al., 2005; Zhao et al., 2006). GBF1 and BIGs localize to cis- and trans- compartments of the Golgi complex, respectively, where they have been proposed to facilitate recruitment of the COPI coat and clathrin adapters (Claude et al., 1999; Kawamoto et al., 2002; Shinotsuka et al., 2002a; Zhao et al., 2002, 2006; Garcia-Mata et al., 2003).

Cell-free assays with Saccharomyces cerevisiae extracts unambiguously established that Sar1-dependent recruitment of COPII drives cargo sorting, as well as budding and release of carriers targeted to the Golgi complex (Barlowe, 2003; Fromme and Schekman, 2005). No such general agreement over the mechanism of cargo export from the ER exists for animal cells, however. In these cells, treatment with BFA or expression of a GDP-arrested Arf mutant blocks export of anterograde cargo from the ER and interferes with its concentration at ERES (Ward et al., 2001; Barzilay et al., 2005). Such observations suggest that formation and release of Golgi carriers in animal cells is more complex and likely involves a two-step process that depends on sequential action of both Sar1 and Arfs (Garcia-Mata et al., 2003; Altan-Bonnet et al., 2004). In this two-step model, Sar1 initially recruits COPII, concentrates cargo, and organizes ER export domains by recruiting additional peripheral proteins such SNAREs and rab1 and its effector p115 (Moyer et al., 2001; Weide et al., 2001). Subsequently, recruitment of GBF1, possibly through its interaction with p115 (Garcia-Mata and Sztul, 2003) leads to Arf activation and the recruitment of numerous effectors that will mature the ER export domains into ERGIC membranes before their release as separate carriers bound for the Golgi complex (Altan-Bonnet et al., 2004). This model is consistent with the observation that expression of a dominant negative mutant of rab1b causes dispersal of the Golgi as observed with BFA, possibly by preventing sequential recruitment of p115 and GBF1 (Alvarez et al., 2003). The two-step model is further supported by the fact that several enteroviral 3A proteins that target GBF1 and block Arf activation, also prevent export from the ER and trap cargo in ERES rather than peripheral VTCs (Wessels et al., 2006a,b).

Here, we report complementary overexpression and (KD) knockdown studies that probe in more detail the role of GBF1 during cargo export from the ER and examine the relative functions of the two Golgi-localized Arf-GEFs in the maintenance and function of the Golgi complex.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Tissue Culture and Reagents
Media and culture reagents were purchased from Invitrogen (Carlsbad, CA). Disposable plastic ware and culture six-well plates were purchased from Falcon Plastics (Oxnard, CA). Monolayers of HeLa, A-549, and BHK-21 cells were maintained in DMEM supplemented with 10% fetal bovine serum (Sigma-Aldrich, St. Louis, MO), 100 µg/ml penicillin G, 100 µg/ml streptomycin, and 2 mM glutamine. For incubations at 32°C, cells were shifted into CO₂-independent media (Invitrogen, Great Island, NY). For the experiment described in Figures 6 and 8, we switched to HeLa cells obtained from ECACC (Sigma-Aldrich, 93021013) were used because they displayed better morphology and clear separation of cis- and trans-Golgi compartments. BFA and monensin were purchased from Sigma-Aldrich (St. Louis, MO). FuGENE 6 and complete protease inhibitor cocktail tablets were purchased from Roche Diagnostics (Indianapolis, IN). All other chemicals were reagent grade and unless otherwise indicated were purchased from Sigma-Aldrich.

Antibodies
9D4 serum was raised against Sec7d-encompassing recombinant proteins containing residues 513–873 of human GBF1. Sera raised against GBF1 (9D2; 1:200) and BIG1 (9D3; 1:500) were described previously (Zhao et al., 2002; Claude et al., 2003). The following additional antibodies were used: anti-BIG2 (a kind gift from Dr. K. Nakayama, Kyoto University, Kyoto, Japan) at 1:100; anti-GFP (Dr. G. Eitzen, University of Alberta, Edmonton, AB, Canada) at 1:2000; anti-AP-1 (clone 88, BD Biosciences Pharmigen) at 1:600; anti-TGN46 (AbD Serotec, Kidlington, Oxford, United Kingdom) at 1:500. Only for immunoblot analysis we used anti-BIG2 (Bethyl Laboratories, Montgomery, TX) and anti-calnexin (Stressgen Biotechnologies, Victoria, BC, Canada). Other antibodies raised against Sec31, giantin, p115 (7D1), ManII, ERGIC-53 (G1/95), and βCOP (M3A5) were used as previously described (Claude et al., 1999; Zhao et al., 2006). Secondary antibodies for this study were as follows: ALEXA488-, ALEXA594-, or ALEXA660-conjugated goat or donkey anti-rabbit and anti-mouse antibodies, as well as Alexa555-conjugated donkey anti-sheep antibody (Molecular Probes, Eugene, OR) at 1:600.

Preparation of Cell Extracts and Analysis by RT-PCR and Immunoblots
For RT-PCR analysis, 1–2 x 10⁶ cells grown in each well of a six-well plate were trypsinized, washed, and processed using the RNAeasy kit according to manufacturer's instructions (Qiagen, Chatsworth, CA). Recovered RNA was analyzed using Qiagen one-step RT-PCR kit according to manufacturer's instructions using gene-specific primers. PCR conditions and cycle numbers were optimized for each primer pair to yield single products of expected size whose level varied in proportion to the amount of RNA added.

Immunoblot analysis of detergent extracts prepared from control and KD cells were carried out essentially as described previously (Zhao et al., 2006). For determination of GBF1 and BIGs KD efficiency and specificity, HeLa cells transfected as described with the appropriate amounts of small interfering RNAs (siRNAs) were lysed after 72 h. Seventy-five-micrograms of each sample were transferred to a nitrocellulose membrane and probed with antisera for GBF1 (9D4, 1:2500), BIG1 (9D3, 1:1000), BIG2 (1:1000), TGN46 (1:2000), AP-1 (1:5000), calnexin (1:20000), β-COP (M3A5, 1:3000), and GGA3 (1:5000) and detected with HRP-conjugated goat anti-rabbit (Bio-Rad Laboratories, Hercules, CA).

Overexpression of VSVG-tsO45 and Other cDNAs
The vesicular stomatitis virus glycoprotein (VSVG)-green fluorescent protein (GFP)-encoding plasmid was a kind gift from Dr. John F. Presley (McGill University, Montreal, QC, Canada). A VSVG-tsO45 virus stock was obtained from Dr. William Balch (Scripps Institute, La Jolla, CA) and grown into a working stock by infection of BHK cells at low multiplicity of infection. Measurement of VSVG traffic in siRNA-treated cells involved separate transfection steps and various combinations of temperature shifts, as illustrated in Figure 3. Briefly, HeLa cells plated at 15% confluency were transfected with the appropriate siRNA duplexes 24 h later. For VSVG-tsO45-GFP expression, KD cells were washed 50 h after RNA transfection, transfected again with 1 µg of VSVG-encoding plasmid, and returned to a 37°C incubator for a further 18 h. Transfected cells were then transferred to a 40°C water-jacketed CO₂ incubator for 4 h, followed by shift to the permissive temperature 32°C for various lengths of time. This shortened incubation at 40°C minimized cellular stress and proved sufficient to accumulate newly synthesized VSVG-protein in the ER. The 32°C incubation was performed in a water bath and required the use of CO₂ independent media. Cells were fixed at different time points after shift to 32°C, as specified. For experiments shown in Figure 3B, cells were transferred directly from 37 to 32°C to bypass the temperature shift to 40°C. For experiments involving live virus, cells were infected with VSV-tsO45 1 h before being shifted to 40°C as described (Zhao et al., 2006). VSVG-GFP was detected either directly using intrinsic GFP fluorescence or by immunofluorescence (IF) using a combination of antibody raised against GFP and ALEXA488-conjugated secondary antibody. The latter method yielded stronger and more stable signal that permitted analysis of cells with low to moderate levels of VSVG in order to avoid artifacts due to overloading the ER with unfolded proteins. Both methods of detection gave similar results. Experiments involving live virus or plasmid-driven expression of VSVG-GFP yielded identical results.

For the monensin treatment experiments, HeLa cells grown to 60% density were transfected with 2 µg plasmid encoding GalT-GFP per 60-mm plate. After 24 h, cells were replated on glass coverslips at 15% confluency and transfected with the appropriate siRNA duplexes 24 h later. Seventy-two hours after siRNA transfection, cells were treated with either 4 µM monensin or equivalent volume of methanol for the periods of time specified and then fixed and processed for IF.

For Arf-GEF overexpression studies (Supplementary Figure 1), BHK cells grown on glass coverslips to 50% density were transfected with 1 µg of purified pCEP4 vector plasmid encoding either GBF1 (Claude et al., 1999) or BIG1 (Mansour et al., 1999), using FuGENE 6 according to the manufacturer's instructions. For exogenous expression of Golgi markers, HeLa cells were cotransfected with 1 µg of plasmid encoding HA-furin that was obtained Dr. J. Bonifacino (Cell Biology and Metabolism Branch, NIH, Bethesda, MD). Twenty-four hours after transfection, cells were treated with 10 µM BFA or an equivalent volume of DMSO, fixed, and processed for IF using the indicated antibodies.

Construction of the Anti-GBF1, -BIG1, and -BIG2 short hairpin RNA–expressing pSUPER-tet Plasmids
To create an inducible form of pSUPER (OligoEngine, Seattle, WA, USA) under the control of the tet repressor, we replaced the promoter sequence of this plasmid with the promoter of the pTER plasmid (Clontech, Mountain View, CA). Briefly, a 2498-base pair fragment between the HindIII and BamH1 sites from pTER was first amplified by PCR and cloned into the pGEM-T Easy vector (Promega, Madison, WI) using overhanging 3' deoxyadenine residues. A 1537-base pair fragment containing the promoter with the tet repressor sequence was then liberated with HindIII and AflII and used to replace the pSUPER promoter using the same restriction sites. The resulting pSUPER-tet plasmid was verified by sequencing.

To generate Arf-GEF-targeting plasmids, pSUPER-tet was linearized by double digestion with HindIII and BglII and synthetic 60-base pair oligomers encoding the desired short hairpin RNA (shRNA) sequences were ultimately inserted using these restriction sites. The double-stranded oligo sequences were designed with cohesive BglII and HindIII sites: (5') GATCCCCTTCAAGAGAYYYYYYYYYYYYYYYYYYYTTTTTA (3') where the 19-nucleotide X and Y residues corresponded to the target sense and antisense sequences identified by Dharmacon (Lafayette, CO) as likely candidates for siRNA-mediated silencing of GBF1, BIG1, and BIG2 expression. All constructions were verified by sequencing. Plasmids were transfected into HeLa cells using 1 µg of plasmid per well of a six-well dish, and cells were fixed after 96 h to allow an extra 24 h for expression and processing of shRNAs.

siRNA Methods
Pools and individual siRNAs targeting different regions of human (h) GBF1 (MU-019783), hBIG1 (MU-012207), hBIG2 (MQ-012208), and hβ-COP (MQ-017940) were purchased from Dharmacon. We followed the Oligofectamine (Invitrogen) transfection protocol for HeLa cells as described (Harborth et al., 2001). Different combinations of targeting duplexes, time points, and siRNA concentrations were assessed to optimize conditions for most effective KD. RT-PCR of cell extracts established that Arf-GEF mRNA levels dropped below 10% by 30 h and remained knocked down for as long as 72 h. Immunoblot analysis of cell extracts was consistent with effective and specific KD (Supplementary Figure 2). Interestingly, IF analysis showed that GBF1 was effectively knocked down in 50–60% of the cells after 72 h after siRNA transfection; in contrast, BIG1 appeared less stable and was undetectable by IF in greater than 70% of the cells as early as 48 h after transfection.

For the experiments presented here, HeLa cells were incubated with a pool of siRNAs targeting sequences 2 and 3 for GBF1, each at a concentration of 100 nM. For BIGs KD, we used a pool of siRNAs targeting sequences 2 and 3 for BIG1 (75 nM each) and 1–4 for BIG2 (50 nM each). For β-COP KD we used sequence 2 at 200 nM. Note that lower siRNA concentrations (50 nM instead of 75 nM for BIG1 and 25 nM instead of 50 nM for BIG2) were sufficient for effective BIGs KD (loss of BIG1 and redistribution of AP-1). As control, cells were exposed to matching concentrations (200–300 nM) of a nonspecific GL2 luciferase siRNA designed as described (Elbashir et al., 2002).

IF Microscopy and Image Analysis
Cells grown on glass coverslips were washed in phosphate-buffered saline (PBS) and fixed with 3% paraformaldehyde in PBS at room temperature for 20 min. Labeling of cells with different antibodies was carried out as described previously (Zhao et al., 2002). Unless otherwise indicated, images were obtained by standard IF with an Axioskop II microscope (Carl Zeiss, Thornwood, NY) using a 63x objective (Plan-APOCHROMAT; NA 1.4) and equipped with a CoolSNAP HQ Photometrics camera (Tucson, AZ). Confocal images (see Figures 4, 6, and 8) were acquired using an LSM 510 (Carl Zeiss) equipped with a similar 63x objective. When two markers were imaged in the same cells, each fluorophore was excited and detected sequentially (multitrack mode) to avoid channel bleed-through. Laser intensity and filters were adjusted to give maximum signal and avoid saturation. Unless otherwise indicated, a single focal plane (0.8–1 µm) was analyzed.

For Figure 8, the line profile analysis was performed using Image-ProPlus Software (Media Cybernetics, Silver Spring, MD). To examine signal distribution within a single stack, we selected red and green structures of similar intensity that clearly appeared contiguous and in close proximity (less then 0.8 µm). Quantification of signal overlap was performed using MetaMorph (Universal Imaging, West Chester, PA).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
GBF1 and BIG1 Regulate the Recruitment of Different Coat Proteins on the Golgi Complex
Several overexpression studies have largely supported functional and selective links between the cis-Golgi localized GBF1 and the COPI coat and on the other hand the trans-Golgi localized BIGs and the clathrin adapter AP-1. For example, Nakayama and colleagues established that overexpression of GBF1 protects Arfs and COPI against BFA-induced redistribution (Kawamoto et al., 2002). This protective effect appears specific because gross overexpression of GBF1 did not prevent release of the clathrin adapter AP-1 (Supplementary Figure 1B), whereas even moderate GBF1 overexpression was sufficient to abrogate the effects of BFA on COPI and the medial Golgi marker Man II (Supplementary Figure 1, A and C). As expected from this specificity, overexpression of GBF1, although it protects the Golgi stack (Claude et al., 1999; Kawamoto et al., 2002; Niu et al., 2005; Zhao et al., 2006; see also Supplementary Figure 1C), did not prevent redistribution of the TGN detected with exogenous hemagglutinin (HA)-furin (Supplementary Figure 1D).

Overexpression of BIGs has effects opposite to GBF1 on stabilization of COPI and clathrin adaptors. Previous work demonstrated that BIG2 overexpression prevented release of the clathrin adaptor AP-1 but not that of COPI (Shinotsuka et al., 2002b). Similarly, BIG1 overexpression had no protective effect on COPI (Supplementary Figure 1A) and the Golgi stack (Supplementary Figure 1C), but did prevent the effects of BFA on the membrane recruitment of the clathrin adaptor AP-1 (Supplementary Figure 1B). Altogether, these results confirm that the two Arf-GEF subfamilies regulate the recruitment of distinct coat proteins on the Golgi complex.

Knockdown of GBF1 Confirms Its Role in Regulating Assembly of the COPI Coat
To examine in more detail the relative function of GBF1 and BIGs in the Golgi complex, we turned to complementary siRNA-based methods to knockdown their expression. RT-PCR analysis established that pools of RNA duplex oligonucleotides targeted to GBF1, BIG1, or BIG2 effectively and selectively knocked down mRNA levels by more than 90% within 30 h of transfection and maintained low mRNA levels for at least 72 h (unpublished data). Analysis of cell extracts by immunoblotting established that GBF1 KD efficiently reduced GBF1 levels with no detectable effect on either β-COP, calnexin, BIGs, or other the TGN markers examined. Similarly, BIGs KD reduced BIGs levels with no detectable effect on AP-1, GGA3, TGN46, β-COP, GBF1, or calnexin (Supplementary Figure 2). To further establish specificity of the KD, we individually tested several RNA duplexes targeting different regions of the GEFs. With the exception of BIG1 1, all individual duplexes yielded KD comparable to the pools; nevertheless, two each of the GBF1 (2 and 3), BIG1 (2 and 3) and BIG2 (1 and 3) targeted RNA duplexes appeared most effective (unpublished data) and were selected for further analysis. Vectors encoding both GFP and shRNAs corresponding to those sequences were constructed. Analysis of transfectants readily identified in the GFP channel confirmed that each RNA sequence effectively suppressed expression of the targeted Arf-GEF (Figure 1).

View larger version (54K): [in this window] [in a new window]   Figure 1. Specific and effective KD of GBF1, BIG1, and BIG2 using siRNA sequences expressed from pSUPER vector. HeLa cells were transfected with 1 µg of GFP-encoding pSUPER plasmids modified with a tet repressor regulated cassette driving expression of either nothing (Ctl) or various individual shRNAs targeting GBF1, BIG1, or BIG2 (see Materials and Methods for construction details). After 96 h, cells were fixed and stained for the markers indicated on the left. Outlines mark GFP-positive transformed cells that were identified in the green channel (shown in inset). Images shown are representative of at least three separate experiments. Bar, 20 µm.

View larger version (40K): [in this window] [in a new window]   Figure 2. Knockdown of GBF1 prevents assembly of the Golgi complex. HeLa cells were transfected with an equimolar mixture of GBF1-targeting siRNA duplexes 2 and 3 as described in Materials and Methods. After 72 h, cells were fixed and double-stained for the markers indicated. Top panels show the pattern observed with GBF1/COPI double-staining. The remaining panels display the distribution of various ERES (Sec31), Golgi (ManII, p115, giantin), and TGN (BIG1, TGN46) markers, with the inset showing the pattern obtained with the second marker (GBF1 or COPI). Cells with effective KD (loss/redistribution of GBF1/COPI) were outlined. Images shown are representative of at least three separate experiments. Bar, 20 µm.

GBF1 Knockdown Does Not Prevent Export from the ER But Blocks Cargo in Post-ERES Peripheral VTC Structures
It has been argued that cargo export from the ER requires a functional COPI system (Altan-Bonnet et al., 2004). This conclusion is based on the observation that BFA blocks export of anterograde cargo molecules such as VSVG and largely prevents their concentration at ERES or VTCs (Ward et al., 2001). The identification of KD conditions that effectively reduce GBF1 levels allowed us to test if a functional COPI system and the activity of GBF1 was necessary for cargo traffic out of ERES. To measure cargo traffic out of the ER, we took advantage of a thermosensitive mutant of VSVG (VSVG-tsO45) that can be accumulated in the ER at the nonpermissive temperature (40°C) and released synchronously upon shift to a permissive temperature (32°C). Cells were transfected with a plasmid encoding a GFP-tagged form of VSVG-tsO45 50 h after RNA-transfection with either luciferase- (Mock), or GBF1-targeted RNA duplexes. As illustrated in Figure 3A, cells were shifted to the nonpermissive temperature to accumulate VSVG in the ER and analyzed at various times after a shift to the permissive temperature.

View larger version (26K): [in this window] [in a new window]   Figure 3. GBF1 and/or COPI KD block VSVG traffic and cause cargo accumulation in peripheral puncta. (A and B) HeLa cells were transfected with siRNA duplexes targeting luciferase (Mock), GBF1, and/or βCOP, as indicated on the left. Fifty hours after transfection, cells were transfected again with a plasmid encoding VSVGtsO45-GFP and then temperature-shifted and fixed, as illustrated. For panel A cells were shifted to the nonpermissive temperature for 4 h before the shift to permissive temperature, whereas in B cells were shifted directly to from 37°C to permissive temperature. Cells were double-stained for COPI and GFP. Images reveal the VSVG-GFP pattern observed at the indicated times after shift to permissive temperature. Knockdown was confirmed by redistribution of COPI (not shown). Images shown are representative of at least three separate experiments. Bar, 20 µm. (C) HeLa cells were transfected with siRNA duplexes targeting luciferase (Mock) or GBF1, as indicated on the left. Fifty hours after transfection, cells were transfected again with a plasmid encoding VSVGtsO45-GFP, then temperature shifted, treated with DMSO/(10 µg/ml) BFA, and fixed, as illustrated. Cells were double-stained for COPI and GFP. Images in the bottom panel reveal the GFP-VSVG pattern. Knockdown was confirmed by redistribution of COPI (not shown). Images shown are representative of at least two separate experiments. Bar, 20 µm.

To obtain independent confirmation that the effect of GBF1 KD resulted from lack of COPI recruitment, we tested in parallel the impact of reducing levels of the β-subunit of COPI on Golgi assembly and function. This subunit plays a critical role in several COPI function such as Arf binding (Zhao et al., 1997) and cargo recruitment (Eugster et al., 2004), and its KD was expected to result in the effective loss of COPI activity. As described in Materials and Methods, four β-COP–targeted RNA duplexes were tested for their ability to disperse Golgi markers (unpublished data) and the most effective one selected for further analysis. Treatment with this RNA duplex effectively caused ManII redistribution (unpublished data) and led to similar accumulation of VSVG cargo into peripheral puncta (Figure 3, A and B). To ascertain that export from ERES was not due to residual COPI activity, we reexamined cargo transport in cells subjected to double KD. As shown in Figure 3, A and B (bottom panels), loss of both GBF1 and COPI did not prevent complete clearing of VSVG from the ER and its accumulation in peripheral puncta.

The accumulation of cargo in puncta observed in GBF1 and COPI KD initially appeared inconsistent with the results previously reported with BFA. This apparent discrepancy prompted us to examine the impact of BFA treatment on cargo accumulation in GBF1 KD cells. Interestingly, treatment with BFA before the shift to permissive temperature prevented accumulation of VSV-G in puncta and yielded the previously reported (Ward et al., 2001), largely reticular pattern in both mock and GBF1 KD cells (Figure 3C). The observation that BFA treatment prevents accumulation of cargo in peripheral VTCs, even in GBF1 KD cells, suggests that the drug affects not only GBF1 activity but additional steps critical to cargo export.

Further analysis of the VSVG positive peripheral puncta identified them as post-ERES structures. Images shown in Figure 4(top panels) first confirmed that they lack COPI, as expected. Furthermore, the VSVG-positive peripheral structures stain for ERGIC53, but not for Sec31, and therefore likely correspond to VTCs (Figure 4, middle and bottom panels). Quantitative analysis of these and five similar images confirmed that greater than 87 ± 6% of the VSVG positive puncta (n = 287) also contain ERGIC53 whereas fewer than 15 ± 3% overlapped with Sec31 (n = 161). Altogether, these results demonstrate that cargo sorting from ERES and transport to peripheral VTCs can take place even in the absence of GBF1/COPI.

View larger version (83K): [in this window] [in a new window]   Figure 4. GBF1 KD traps carso in ERGIC53-positive VTCs that are separate from Sec31-positive ERES. HeLa cells were transfected first with siRNA duplexes targeting GBF1 and 50 h later with a plasmid encoding VSVGtsO45-GFP as described in Materials and Methods. After 18 h, cells were shifted to 40°C for 4 h and then to 32°C for 120 min, as illustrated in Figure 3A. Fixed cells were double-stained for the indicated markers. Single-slice confocal images are shown. Insets display threefold magnification of boxed areas of merged images shown in center panels. Images shown are representative of at least two separate experiments. Bar, 20 µm.

To confirm the involvement of BIGs in clathrin adaptor recruitment, we first examined the impact of BIGs KD on the distribution of endogenous AP-1 and GGA3. As shown in Figure 5A, and B, BIGs KD eliminated bright juxta-nuclear staining for AP-1, yielding a weaker and more dispersed punctate pattern. Quantitative analysis of these and similar images from three separate experiments confirmed that BIGs KD caused AP-1 redistribution in greater than 85% of transfected cells (n = 61). BIGs KD similarly caused redistribution of endogenous GGA3 from a compact juxtanuclear structure to a diffuse pattern (Figure 5C). We observed identical AP-1 redistribution whether performing KD treatment for 2 or 3 d, which led us to conclude that residual membrane association of clathrin adapters likely reflects Arf activation by endosome associated Arf-GEFs of the ARNOs and EFA6 subfamilies (D'Souza-Schorey and Chavrier, 2006).

View larger version (37K): [in this window] [in a new window]   Figure 5. BIGs KD blocks recruitment of TGN-specific coats, redistributes TGN markers, but does not prevent assembly the Golgi stack. HeLa cells were transfected with pools of siRNA duplexes targeting both BIG1 (2 and 3) and BIG2 (1–4) as described in Materials and Methods. Fixed cells were double-stained for the indicated markers. (A and B) The pattern observed with BIG1/AP-1 double-staining. (C–H) The distribution of various Golgi (COPI, p115, ManII) and TGN (GGA3, TGN46, sortilin) markers with the inset showing pattern obtained with the second marker (BIG1 or AP-1). Cells with effective KD (loss/redistribution of BIG1/AP-1) were outlined. Images shown are representative of at least three separate experiments. BIGs KD for 3 d did not cause detectable reduction or fragmentation of Golgi signal (Man II, p115, and COPI) in any of the transfectants examined (n = 30, 142, and 39, respectively). Bar, 20 µm.

BIGs KD Disrupts Assembly of the TGN
BIGs KD and/or consequent loss of adaptor recruitment in the Golgi region leads to loss of a detectable TGN. The first indication that BIGs may be essential for TGN assembly came from examination of the well-characterized TGN membrane marker TGN46. Effective BIGs KD, as measured by AP-1 redistribution (inset), caused dispersal of TGN46 (Figure 5G). This marker did not accumulate at the cell surface but rather relocalized to weak puncta scattered throughout the cytoplasm, as previously observed in GBF1 KD cells (Figure 2H). To further probe the impact of BIGs KD, we examined the distribution of cargo receptors that cycle between the TGN and endosomes to transport lysosomal hydrolases and whose function rely on GGAs (Bonifacino and Traub, 2003; Ni et al., 2006). These sorting receptors include not only canonical mannose 6-phosphate receptors, but also the more recently described sortilin, a member of the Vps10p family. Several experiments established that the cation independent mannose 6-phosphate receptor retained the characteristic punctate pattern of a late endosomal marker in BIGs KD cells (unpublished data). In sharp contrast, BIGs KD reproducibly prevented normal localization of sortilin; this marker redistributed to a weak dispersed pattern and did not appear to accumulate at the cell surface (Figure 5H).

To examine whether the redistribution of TGN markers resulted from a defect in assembly and maintenance of a functional TGN, we took advantage of the fact that treatment with the proton ionophore monensin traps a subset of Golgi enzymes such as GalT into dispersed vacuoles derived from the TGN (Borsig et al., 1999; Puri et al., 2002; Schaub et al., 2006). GalT resides primarily in trans-Golgi cisternae but does cycle to and from the TGN where it becomes trapped in monensin-treated cells (Schaub et al., 2006). Any residual trafficking to the TGN in BIGs KD cells should therefore be detectable by accumulation of GalT in remaining TGN structures after treatment with monensin. We first verified that BIGs KD had no impact on the Golgi localization of GalT and its codistribution with the medial Golgi marker ManII (Figure 6, top panels). Second, we confirmed that in control cells a brief 15 min treatment with monensin was sufficient to cause accumulation of GalT in vacuoles clearly separate from the Golgi ribbon (Figure 6, middle panels). As shown in Figure 6 (bottom panels), monensin did not change the GalT juxta-nuclear distribution in BIGs KD cells, even when the treatment was lengthened from 15 to 30 min. Quantitative analysis revealed that GalT redistributed to dispersed vacuoles distinct from ManII positive Golgi ribbons in more than 91 ± 4% of control cells (n = 117), whereas this occurred in fewer than 14 ± 10% of KD cells (n = 46). The lack of GalT redistribution in the majority of BIGs KD cells demonstrates that effective KD was achieved and strongly suggests that BIGs play a critical role in assembly of the TGN.

View larger version (85K): [in this window] [in a new window]   Figure 6. BIGs KD prevents monensin-induced redistribution of GalT-GFP. HeLa cells were transfected with a plasmid encoding GalT-GFP and a pool of siRNA duplexes targeting either luciferase (Mock), or BIG1 and BIG2 (BIGs) as described in Materials and Methods. Seventy-two hours after siRNA transfection, cells were treated with either methanol or 4 µM monensin for the indicated periods of time. Cells were then fixed and stained for ManII and AP-1. Cells with effective BIGs KD were identified by loss of AP-1 juxtanuclear staining. Single-slice confocal images of the ManII (red) and GFP signal are shown. Merged images shown on the right. Images shown are representative of at least three separate experiments. Bar, 20 µm.

View larger version (18K): [in this window] [in a new window]   Figure 7. BIGs KD does not block traffic of VSVG to the cell surface. HeLa cells were transfected with siRNA duplexes targeting either luciferase (Mock) or BIG1 and BIG2 (BIGs), as indicated on the left. Fifty hours after transfection, cells were transfected again with a plasmid encoding VSVGtsO45-GFP, then temperature shifted, and fixed, as illustrated in Figure 3A. (A) Cells fixed at the indicated times after shift-down to 32°C were double-stained for BIG1 and GFP. Knockdown was confirmed by redistribution of BIG1 (not shown). Images reveal the GFP-VSVG pattern observed at the times shown above the panels. Arrowheads mark plasma membrane (PM) where VSVG accumulates. Cells with effective KD (loss of BIG1) were marked by asterisks. Images shown are representative of at least six separate experiments. Bar, 20 µm. (B) At each time point after release for both mock ( and , solid line) and BIGs KD ( and , dashed line) cells, a minimum of 25 cells were scored for presence of VSVG at Golgi only (, ), or at Golgi and PM (, ). The fraction of cells with the indicated pattern was expressed as percentage and is shown as a function of time after shift-down to 32°C.

To better illustrate the spatial resolution of the Golgi markers, we measured signal intensity along the white line shown in the merged image and reported values for each marker in the graphs on the right. The graphs confirmed good overlap of BIG1 and AP-1 with GalT-GFP in mock-treated cells and clear separation of cis- and trans-markers in both mock-treated and BIGs KD cells. Quantitative analysis of the average distance between cis- and trans-Golgi markers in several regions (n > 6) of cells from two separate experiments established that the average distance between p115/GalT or GBF1/GalT peaks were nearly identical in control and BIGs KD cells. Furthermore, quantitative analysis of fluorescence signal overlap in several images similar to those shown in Figure 8 confirmed that both pairs of cis/trans markers remained well resolved in BIGs KD cells (Table 1). Altogether, these results strongly suggest that GBF1/COPI, but not BIGs/clathrin, is essential to drive assembly and maintenance of a polarized Golgi stack.

View larger version (63K): [in this window] [in a new window]   Figure 8. BIGs KD does not prevent assembly of a polarized Golgi stack. HeLa cells were transfected with a pool of siRNA duplexes targeting either luciferase (Mock), or BIG1 and BIG2 (BIGs). After 48 h, cells were transfected again with a plasmid encoding GalT-GFP. Cells were fixed 24 h later and double-stained for either BIG1/p115 or AP-1/GBF1. Single-slice confocal images are shown. Merged images display the red and GFP signal from the enlarged boxed areas. Graphs in the right column report pixel intensity profiles in all three channels along the white bar shown in the merged panels. Images shown are representative of at least three separate experiments. Bar, 20 µm.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Effectiveness and Selectivity of Golgi ArfGEFs KD
Several experiments established that the KD methods used in our study led to selective and effective loss of the targeted proteins. Multiple observations confirmed the selectivity of the KD effects. These include the fact that similar KD effects were observed using any one of several sequences targeting different regions of the mRNA, and whether sequences were delivered by direct transfection of duplexes or by plasmid-driven synthesis of short hairpins RNA. The observation that targeting either β-COP or GBF1 had the same impact on ER export and the Golgi complex further supports our conclusion that KDs were selective.

The extent of KD varied within any given cell population, and depended on the nature of the target and length of treatment. These variations allowed us to identify with confidence cells displaying effective KD of targeted proteins. For example, we could readily recognize cells with partial KD for GBF1 at shorter treatment times (48 h); these cells lacked or showed little detectable GBF1 but still displayed a fragmented Golgi complex stained weakly with COPI. By lengthening treatment time to 72 h and selecting cells with no remaining COPI staining, we could ensure that all cells analyzed had effective GBF1 KD. In the case of BIGs, short treatment (48 h) proved effective at eliminating BIG and dispersing AP-1 IF signal. The fact that lengthening treatment to 72 h had no further impact on the distribution of TGN and Golgi markers established that effective KD had been achieved.

GBF1 Is Essential for COPI Recruitment and Assembly of the Golgi Complex, But Is Not Required for Cargo Concentration and Export from ERES
Results obtained through our combined use of overexpression and RNA-dependent silencing established that GBF1, but not BIGs, is required to activate Arfs for recruitment of COPI. The localization of GBF1 to early compartments of the Golgi complex first suggested a functional link between GBF1 and COPI (Kawamoto et al., 2002; Zhao et al., 2002). Subsequent studies revealed that overexpression of the charge-reversal dominant negative mutant GBF1[E794K] (Garcia-Mata et al., 2003) or microinjection of neutralizing GBF1 antibodies (Zhao et al., 2006) caused loss of COPI recruitment. The loss of COPI recruitment and Golgi structure after GBF1 KD (Figure 2), as well as the coat and compartment-specific protection conferred by GBF1 overexpression (Supplementary Figure 1), extend these observations and confirm this functional link.

The availability of tools for the effective KD of GBF1 and COPI allowed us to test the two-step model which proposes that Arfs and COPI are required for traffic out of ERES. Contrary to predictions from this model, we observed that the GBF1/COPI machinery was not required for the concentration and export of VSVG cargo from ERES. These results are consistent with a wealth of information derived from yeast cell-free assays (Barlowe, 2003; Sato and Nakano, 2007) and the recent identification of COPII carriers in animal cells (Zeuschner et al., 2006). These results are also consistent with our previous demonstration that GBF1 does not associate with ERES, but rather with VTCs that are close but physically separate from ERES (Zhao et al., 2006).

To explain the apparent block in cargo export by BFA or Arf mutants (Ward et al., 2001), we propose that these treatments do not block ER export, but rather prevent their retention in VTCs by promoting retrograde traffic from VTCs to ER. This model is based on our recent demonstration that BFA causes VTCs to lose their cargo to the ER through a microtubule-dependent mechanism (Zhao et al., 2006). This could occur if the inactive GBF1 trapped on membranes by BFA (Niu et al., 2005; Szul et al., 2005; Zhao et al., 2006) somehow interfered with protein sorting at ERES and/or VTCs. Alternatively, BFA could affect additional targets such as BARS, which has been implicated in membrane scission of COPI vesicles (Yang et al., 2005) and whose activity is inhibited by BFA-induced ADP-ribosylation (Weigert et al., 1999).

As initially suggested by the impact of the GBF1[E794K] mutant (Garcia-Mata et al., 2003), GBF1 KD prevented the formation of motile transport competent carriers necessary for assembly and maintenance of the Golgi complex. In GBF1 KD cells, VSVG cargo accumulated in VTCs that contained several tethering factors such as p115 and giantin, but failed to either mature into Golgi resident enzyme-containing structures or to associate with microtubules and migrate to the cell center. The fact that VSVG but not Golgi resident enzymes accumulated in VTCs likely reflects the presence in VSVG of a di-acidic sorting signal efficiently recognized by COPII (Nishimura et al., 1999; Sato and Nakano, 2007). The reason for the lack of movement to the cell center remains unknown but, as proposed by Sztul and colleagues (Garcia-Mata et al., 2003), it may be related to the absence of COPI-driven active protein sorting that normally drives formation of membrane domains critical for recruitment of other proteins such as rabs and motors/accessory proteins (Short et al., 2005).

BIGs Are Required for Recruitment of Clathrin Adaptor and Maintenance of the TGN
Previous work established that BIGs localize at the TGN and overlap with clathrin, suggesting that BIGs regulate Arf activation for recruitment of GGAs and other adaptor proteins such as AP-1 (Yamaji et al., 2000; Zhao et al., 2002). We confirmed here the functional link between BIG1 and AP-1 by showing that BIG1 overexpression stabilizes AP-1 but not COPI against dispersal after short BFA treatment. Nakayama and colleagues confirmed a similar link between BIG2 and AP-1 using related approaches (Shinotsuka et al., 2002b). As predicted, BIGs KD caused loss of both AP-1 and GGA3 from the juxta-nuclear region. This redistribution of AP-1 is similar to that reported after BFA treatment of several cell lines in which the Golgi complex is either naturally resistant to BFA or acquired resistance after mutagenesis. For example, the Golgi stack of MDCK and PtK1 cells (Robinson and Kreis, 1992) or mutagenized CHO-K1 cells (Torii et al., 1995) remains unperturbed after BFA treatment, whereas some AP-1 localizes to disperse puncta. In all cases, residual membrane association of clathrin adapters likely results from Arf activation by endosome-associated BFA resistant Arf-GEFs of the ARNO and EFA6 subfamilies (D'Souza-Schorey and Chavrier, 2006).

Previous attempts to test the model that GGAs and APs recognize sorting signals in endosomal-targeted cargo and drive maturation of the TGN, focused on blocking the function of either AP1, -3, and -4 or GGAs (Bonifacino and Traub, 2003; Gleeson et al., 2004). However, expression of dominant negative mutants or silencing of either types of adaptors led to variable outcomes ranging from tubulation of the Golgi, accumulation of endosomal cargo in the TGN or its dispersal to peripheral endosome structures (Puertollano et al., 2001a,b; Ghosh et al., 2003). These apparent discrepancies should not be surprising, however, because current evidence suggests that adaptors function in both anterograde and retrograde traffic and at the TGN participate in a multistep process involving both GGAs and APs (Ghosh and Kornfeld, 2004); loss of only a subset of the adaptors would imbalance this process, with complex consequences. Effective KD of BIGs circumvented this problem by preventing recruitment of both APs and GGAs to the Golgi complex. Under these conditions, we could readily detect loss of a recognizable juxta-nuclear TGN structure stained by TGN46 or sortilin. Disruption of sorting to the TGN was confirmed by clear loss of GalT accumulation in dispersed vacuoles after monensin treatment. We predict that simultaneous silencing of all GGAs and several APs will be required to observe effects similar to those we report for BIGs KD.

GBF1, But Not BIGs, Is Required for Assembly of a Polarized Golgi Stack
One of the more unexpected results of our study was the observation that BIGs KD did not prevent assembly of a polarized Golgi stack in mammalian cells; these stacks not only retained a degree of polarization similar to that of control cells, but also efficiently trafficked VSVG to the cell surface (Figures 7 and 8). These observations were surprising because previous work in S. cerevisiae had established that loss of Sec7p, the single orthologue of BIGs, completely alters Golgi morphology and blocks protein secretion. For example, Sec7 temperature-sensitive mutants accumulate large numbers of stacked Golgi membranes (Berkeley bodies) with concomitant block in traffic to the vacuole and cell surface (Novick et al., 1980; Esmon et al., 1981; Franzusoff and Schekman, 1989; Rambourg et al., 1993; Deitz et al., 2000). Although undetectable levels of remaining BIGs could account for those observations, we consider this possibility extremely unlikely as argued above. We propose instead that the apparent discrepancy between the impact of Sec7 inactivation and BIGs KD may reflect differences in the organization of the secretory pathways of S. cerevisiae and animals.

Yeast Golgi elements appear as fine or coarse nodular networks that are neither cisternal nor arranged in stacks (Morin-Ganet et al., 2000; Rambourg et al., 2001; Kepes et al., 2005). These morphogenetic studies revealed a gradual transformation of Golgi elements after their initial assembly from ER-derived vesicles: from tubular clusters, they become a network of fine tubules linked by nodes that transform into a thicker nodular network that eventually releases secretory granules by rupture of tubular areas. The presence of mixed forms containing two networks of different calibers (Morin-Ganet et al., 2000) actually suggests that the gradual transformation from early to late elements involves concerted action of both the COPI and clathrin coats within a continuous network. This view is consistent with recent description of Golgi maturation in live yeast (Losev et al., 2006; Matsuura-Tokita et al., 2006) and may explain why loss of Sec7 would prevent maturation of the nodular network and cause accumulation of membranes unable to release cargo carriers.

In contrast, the Golgi complex of animal cells occurs as a structured stack of discrete flattened cisternae with varying extent of fenestration that is flanked by extensive tubular-reticular networks (Rambourg and Clermont, 1990; Thorne-Tjomsland et al., 1998; Mogelsvang et al., 2004). Our results suggest that the GBF1-dependent recruitment of the COPI coat is sufficient to promote the formation of specialized membrane domains and cargo carriers that can move cargo from VTCs, assemble onto the Golgi matrix and subsequently drive the maturation process to yield a polarized stack. The fact that one can eliminate BIGs and maintain much of this organization suggests that these mechanisms are robust and that stacking of separate cisternae may permit the observed uncoupling of cis- and trans-acting coats.

The observation that VSVG trafficked normally to the cell surface in BIGs KD cells remains consistent with the well-established role of the TGN in sorting of cargo to various destinations in both nonpolarized and polarized cells (Bonifacino and Traub, 2003; Rodriguez-Boulan et al., 2005; Rodriguez-Boulan and Musch, 2005; Bard and Malhotra, 2006). Maturation by the COPI coat in animal cells may allow the creation of membrane domains on the trans-side that become enriched in anterograde cargo proteins and eventually peel off the stack to carry some or all of its content to the cell surface. Under normal conditions such intermediates would likely be absorbed in the TGN from which cargo would be then sorted to its various destinations. Our results establish that whatever mechanisms normally drive traffic of VSVG from the TGN remain operational in the absence of BIGs/clathrin-dependent sorting. The absence of a detectable TGN in BIGs KD cells raises one additional issue: why so few, if any, of the membranes released from the Golgi stack remain in this area. Further work will clearly be required to establish when and how the BIGs machinery is recruited to late cisternae to facilitate membrane retention, formation of the TGN, and sorting of endosomal cargo.

	ACKNOWLEDGMENTS

 
We thank L. Channon for maintenance of cultured cells and RT-PCR analysis, Z. Shapovalova for construction of pSUPER vectors targeting BIG2, and F. Mast for advice on image quantitation as well as H. Chan for technical help with confocal microscopy. We thank Drs. K. Nakayama and H.-W. Shin (Kyoto University) for the BIG2 antibody and Dr. J. Rohrer (University of Zurich) for advice on the monensin experiments. Finally, we thank Drs. T. Hobman and Z. Wang (University of Alberta) for helpful discussions. This study was initiated with funds from the Alberta Heritage Foundation for Medical Research and supported by grants (P.M.) from the Canadian Institutes of Health Research and the Human Frontier Science Program. A.C. and J.R. were supported by the Fondecyt Grant 1030346.

	Footnotes

 
This was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E07-04-0394) on November 14, 2007.

Address correspondence to: Paul Melançon (Paul.Melancon{at}UAlberta.ca)

Abbreviations used: AP, adaptor protein; Arf, ADP-ribosylation factor; BFA, brefeldin A; BIG, brefeldin A–inhibited guanine nucleotide exchange factor; COP, coat protein; ERES, ER exit sites; ERGIC, ER-Golgi intermediate compartment; GBF, Golgi-specific brefeldin A resistance factor; GEF, guanine nucleotide exchange factor; GGA, gamma ear Golgi-localized Arf-binding protein; IF, immunofluorescence; ManII, mannosidase II; KD, knockdown; PM, plasma membrane; shRNA, short hairpin RNA; siRNA, small inhibitory RNA; TGN, trans-Golgi Network; VSVG, vesicular stomatitis virus glycoprotein; VTC, vesiculotubular cluster.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Altan-Bonnet, N., Sougrat, R., and Lippincott-Schwartz, J. (2004). Molecular basis for Golgi maintenance and biogenesis. Curr. Opin. Cell Biol 16, 364–372.[CrossRef][Medline]

Alvarez, C., Garcia-Mata, R., Brandon, E., and Sztul, E. (2003). COPI recruitment is modulated by a Rab1b-dependent mechanism. Mol. Biol. Cell 14, 2116–2127.[Abstract/Free Full Text]

Appenzeller-Herzog, C., and Hauri, H. P. (2006). The ER-Golgi intermediate compartment (ERGIC): in search of its identity and function. J. Cell Sci 119, 2173–2183.[Abstract/Free Full Text]

Bard, F., and Malhotra, V. (2006). The formation of TGN-to-plasma-membrane transport carriers. Annu. Rev. Cell Dev. Biol 22, 439–455.[CrossRef][Medline]

Barlowe, C. (2003). Signals for COPII-dependent export from the ER: what's the ticket out? Trends Cell Biol 13, 295–300.[CrossRef][Medline]

Barzilay, E., Ben-Califa, N., Hirschberg, K., and Neumann, D. (2005). Uncoupling of brefeldin a-mediated coatomer protein complex-I dissociation from Golgi redistribution. Traffic 6, 794–802.[CrossRef][Medline]

Bonifacino, J. S. (2004). The GGA proteins: adaptors on the move. Nat. Rev. Mol. Cell Biol 5, 23–32.[CrossRef][Medline]

Bonifacino, J. S., and Glick, B. S. (2004). The mechanisms of vesicle budding and fusion. Cell 116, 153–166.[CrossRef][Medline]

Bonifacino, J. S., and Traub, L. M. (2003). Signals for sorting of transmembrane proteins to endosomes and lysosomes. Annu. Rev. Biochem 72, 395–447.[CrossRef][Medline]

Borsig, L., Imbach, T., Hochli, M., and Berger, E. G. (1999). alpha1,3Fucosyltransferase VI is expressed in HepG2 cells and codistributed with beta1,4galactosyltransferase I in the golgi apparatus and monensin-induced swollen vesicles. Glycobiology 9, 1273–1280.[Abstract/Free Full Text]

Claude, A., Zhao, B. P., Kuziemsky, C. E., Dahan, S., Berger, S. J., Yan, J. P., Armold, A. D., Sullivan, E. M., and Melancon, P. (1999). GBF1, a novel Golgi-associated BFA-resistant guanine nucleotide exchange factor that displays specificity for ADP-ribosylation factor 5. J. Cell Biol 146, 71–84.[Abstract/Free Full Text]

Claude, A., Zhao, B. P., and Melancon, P. (2003). Characterization of alternatively spliced and truncated forms of the Arf guanine nucleotide exchange factor GBF1 defines regions important for activity. Biochem. Biophys. Res. Commun 303, 160–169.[CrossRef][Medline]

Cox, R., Mason-Gamer, R. J., Jackson, C. L., and Segev, N. (2004). Phylogenetic analysis of Sec7-domain-containing Arf nucleotide exchangers. Mol. Biol. Cell 15, 1487–1505.[Abstract/Free Full Text]

D'Souza-Schorey, C., and Chavrier, P. (2006). ARF proteins: rol